Chinese Textual Retrieval Based on Fuzzy Concept Network

Abstract

In this paper, a Chinese document retrieval system based on a fuzzy concept network is proposed and implemented. The system contains two main parts, namely the network module and the query module. The proposed network is realized with concept matrices so that searching through the network is simplified to be a matrix computation. Meanwhile, the relation values among network nodes are computed by proposed weighting functions which take keywrod occurrence freauencies and location information into account. On the other hand, the query module is incorporated with a proposed concept -base relevant document retrieval module in order to facilitate relevance feedback. Finally the system is tested with a collection of real Chinese technical documents and queries. The experimental results show that the proposed concept-based retrieval yield higher retrieval accuracy rate than the traditional simple keyword-matching retrieval.